Authority, Refusal, and Resilience in Autonomous Systems

Authority, Refusal, and Resilience in Autonomous Systems develops the Stable Authority Boundary (SAB) as a pre-execution design invariant for autonomous, distributed, AI-enabled, and safety-critical systems. The monograph argues that consequential execution is legitimate only when the applicable authority is identified, current, valid, within scope, and verifiable before action begins. Where authority is absent, expired, revoked, ambiguous, degraded, out of scope, or otherwise unverifiable, the correct engineered response is refusal, hold, silence, or another explicitly governed non-action state. The work treats authority, refusal, silence, and resilience as execution-level engineering properties rather than post-hoc governance abstractions. It examines authority contraction under uncertainty, refusal as a legitimacy-preserving enforcement mechanism, the misdiagnosis of silence, override and authority drift, degraded coordination, and resilience as preservation of legitimate purpose rather than uptime alone. A central concern is the evidence on which autonomous execution depends. Modern systems increasingly consume information from external repositories, sensors, services, AI pipelines, and other automated processes. That makes machine-readable provenance, dataset provenance, dataset integrity, metadata generation, data validation, automated transformation, and computational reproducibility directly relevant to execution authority. A system may receive an open research dataset, sample dataset, derived dataset, or information maintained within a research data archive. It may encounter a data processing example, execute an automated transformation, or consume a result whose identity is represented through a dataset DOI or data DOI. Technical availability alone does not establish that the information remains valid for the proposed action. SAB therefore treats provenance, validation, integrity, and current authority as related evidence questions. Where a source or transformation cannot be sufficiently established, tamper-evident data, machine-readable records, persistent identifiers, and reproducible processing can help determine what information was used, how it changed, and whether the resulting action remains supportable. This revised edition aligns the monograph with the Stable Authority Boundary (SAB) v1.0 — Conformance Specification, CSI SAB-STD-v1.0-DF-2026, and extends the work with a practical SAB review framework, eight conformance dimensions, evidence-bearing operation, and a hostile falsification audit. Stable Authority Boundary (SAB) v1.0 — Conformance Specificationhttps://doi.org/10.5281/zenodo.21939806 BLOCK VECTOR Research Maphttps://doi.org/10.5281/zenodo.22102620 Interactive BLOCK VECTOR Research Maphttps://blockvectortech.com/research-map.html Author ORCID — David B. Forbeshttps://orcid.org/0009-0003-8833-3853